11.3 Summary

11.3 Summary#

Specifying a Model#

desc = '''# Measurement model
          latent_factor1 =~ x1 + x2 + x3
          latent_factor2 =~ x7 + x8 + x9
          latent_factor3 =~ x4 + x5 + x6
          
          # Addding higher order factors
          latent_factor1 =~ latent_factor2
          latent_factor1 =~ latent_factor3

          # Structural model
          latent_factor1 ~ latent_factor2

          # Adding a covariance
          latent_factor2 ~~ latent_factor3
          
          # Setting a covariance to zero
          latent_factor1 ~~ 0*latent_factor3

          # Setting a factor variance to 1
          latent_factor1 ~~ 1 * latent_factor1'''

Summed up, you can use the following operators:

  • =~ to associate measured variables with latent factors (or latent factors with higher order latent factors)

  • ~ for regressions

  • ~~ for variances and covariances

Fitting a Model#

mod = semopy.Model(desc)
res_opt = mod.fit(data)

Extracting Model Estimates#

estimates = mod.inspect()
print(estimates)

Extracting Model Fit Measures#

stats = semopy.calc_stats(mod)
print(stats.T)

Visualizing the Model#

semopy.semplot(mod, plot_covs = True, filename='data/cfa_plot.pdf')
Contents